haitengzhao
commited on
Commit
•
bb54c34
1
Parent(s):
ef0ee08
Upload 2 files
Browse files- config.json +85 -0
- generation_config.json +7 -0
config.json
ADDED
@@ -0,0 +1,85 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "ckpts/gimlet",
|
3 |
+
"architectures": [
|
4 |
+
"GraphT5TransformerForConditionalGeneration"
|
5 |
+
],
|
6 |
+
"d_ff": 2048,
|
7 |
+
"d_kv": 64,
|
8 |
+
"d_model": 512,
|
9 |
+
"decoder_start_token_id": 0,
|
10 |
+
"dense_act_fn": "relu",
|
11 |
+
"dropout_rate": 0.1,
|
12 |
+
"eos_token_id": 1,
|
13 |
+
"feed_forward_proj": "relu",
|
14 |
+
"graph_args": {
|
15 |
+
"apply_graphormer_init": true,
|
16 |
+
"dropout": 0.0,
|
17 |
+
"edge_type": "multi_hop",
|
18 |
+
"encoder_attention_heads": 8,
|
19 |
+
"encoder_embed_dim": 512,
|
20 |
+
"encoder_layers": 6,
|
21 |
+
"encoder_normalize_before": false,
|
22 |
+
"graphonly_problem_type": "",
|
23 |
+
"graphonly_readout": "mean",
|
24 |
+
"loss_reduction_method": "token",
|
25 |
+
"maskt2g": true,
|
26 |
+
"max_nodes": 128,
|
27 |
+
"multi_hop_max_dist": 5,
|
28 |
+
"num_atoms": 4608,
|
29 |
+
"num_classes": -1,
|
30 |
+
"num_edge_dis": 128,
|
31 |
+
"num_edges": 1536,
|
32 |
+
"num_in_degree": 512,
|
33 |
+
"num_out_degree": 512,
|
34 |
+
"num_spatial": 512,
|
35 |
+
"spatial_pos_max": 1024,
|
36 |
+
"transformer_backbone": "gimlet"
|
37 |
+
},
|
38 |
+
"initializer_factor": 1.0,
|
39 |
+
"is_encoder_decoder": true,
|
40 |
+
"is_gated_act": false,
|
41 |
+
"layer_norm_epsilon": 1e-06,
|
42 |
+
"loss_reduction_method": "token",
|
43 |
+
"model_type": "t5",
|
44 |
+
"n_positions": 512,
|
45 |
+
"num_decoder_layers": 6,
|
46 |
+
"num_heads": 8,
|
47 |
+
"num_layers": 6,
|
48 |
+
"output_past": true,
|
49 |
+
"pad_token_id": 0,
|
50 |
+
"relative_attention_max_distance": 128,
|
51 |
+
"relative_attention_num_buckets": 32,
|
52 |
+
"task_specific_params": {
|
53 |
+
"summarization": {
|
54 |
+
"early_stopping": true,
|
55 |
+
"length_penalty": 2.0,
|
56 |
+
"max_length": 200,
|
57 |
+
"min_length": 30,
|
58 |
+
"no_repeat_ngram_size": 3,
|
59 |
+
"num_beams": 4,
|
60 |
+
"prefix": "summarize: "
|
61 |
+
},
|
62 |
+
"translation_en_to_de": {
|
63 |
+
"early_stopping": true,
|
64 |
+
"max_length": 300,
|
65 |
+
"num_beams": 4,
|
66 |
+
"prefix": "translate English to German: "
|
67 |
+
},
|
68 |
+
"translation_en_to_fr": {
|
69 |
+
"early_stopping": true,
|
70 |
+
"max_length": 300,
|
71 |
+
"num_beams": 4,
|
72 |
+
"prefix": "translate English to French: "
|
73 |
+
},
|
74 |
+
"translation_en_to_ro": {
|
75 |
+
"early_stopping": true,
|
76 |
+
"max_length": 300,
|
77 |
+
"num_beams": 4,
|
78 |
+
"prefix": "translate English to Romanian: "
|
79 |
+
}
|
80 |
+
},
|
81 |
+
"torch_dtype": "float32",
|
82 |
+
"transformers_version": "4.28.0.dev0",
|
83 |
+
"use_cache": true,
|
84 |
+
"vocab_size": 32100
|
85 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"decoder_start_token_id": 0,
|
4 |
+
"eos_token_id": 1,
|
5 |
+
"pad_token_id": 0,
|
6 |
+
"transformers_version": "4.28.0.dev0"
|
7 |
+
}
|